Estimating assortative mating, its history, and its future effect on genetic variance for health, behavioral, and ancestry phenotypes using crosssectionaldata

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT Assortative mating (AM) is when parents are more similar genetically than if pairing were random. AM can lead to a greater prevalence and severity of disease in the population and can mislead researchers studying the relationships between phenotypes. A key objective of this project is to precisely estimate how parents have sorted over time to describe past and future trends in public health—and to show how to use those estimates to take AM into account when trying to narrow down the search for the biological roots of diseases. The key idea behind this project is that AM leads to the next generation having a larger variance of genetic risk—measured by polygenic scores (PGSs)—than it otherwise would. Greater variance leads to a larger range for the PGS and a greater incidence of extreme values. A key concern about AM is that for a disease PGS, a greater incidence of extreme values associated with a larger variance will lead to a greater overall incidence and a greater incidence of more severe forms of the disease. But that increase in variance also provides a valuable tool to study AM. For example, if the variance of a PGS is changing in the population, we can use this information to estimate how AM must also be changing. Genetic data on parent pairs is currently scarce—measuring in the tens of thousands—while cross-sectional data on unrelated individuals is measured in millions. This project is to develop and deploy approaches using cross-sectional data in unrelated individuals to study the dynamics and average level of assortative mating for different individual diseases and traits and across pairs of different diseases and traits. We will validate the results using data on spouse pairs. This project has three specific aims: 1. Develop a general theoretical framework to understand how AM affects the variance and correlation of PGSs across individuals. Importantly, AM can affect the distribution of genetic risk for many generations. 2. Use the theoretical framework to develop methods to estimate AM and its changes using data on unrelated individuals. This will include studying AM for individual phenotypes and AM between pairs of phenotypes. 3. Estimate the history of AM and cross-phenotype AM and study illustrative implications for health and health research, including (a) forecasting the effect of changes in AM on future disease incidence and future disease comorbidities, (b) parceling out the part of past trends in disease incidence and comorbidities that can be accounted for by changes in AM, (c) sorting out what part of the typical level of spousal similarity in disease risk is due to AM and what part is due to other forces, such as a common environment that is generative of that disease, (d) sorting out what portion of the similarity between genes that predict a disease and genes that predict another disease or trait is due to AM and therefore does not suggest a common biological pathway, (e) correcting other crucial genetic measures for AM.